Deep Learning Fluorescence Imaging of Visible to NIR-II Based on Modulated Multimode Emissions Lanthanide Nanocrystals

Fluorescence bioimaging has always been a research hotspot in the field of life sciences and medicine. Although many studies focus on the promising second near infrared window (NIR-II) imaging, the NIR-II imaging with deep tissue penetration is limited by the broad emission band widths. Herein, a well-designed lanthanide doped nanocrystal is presented that can modulate the energy migration processes by controlling energy migration pathway and cerium-assisted energy transfer processes, resulting in switchable emission modes of visible and NIR-II that dependent by the excitation wavelengths. Subsequently, the multimode emissions of dumbbell-like nanocrystals are cooperated with deep learning, where the advantages of narrow emission peak of visible fluorescence and deep tissue penetration of NIR-II fluorescence are combined to offer a unique deep learning fluorescence bioimaging. By this new imaging method, fluorescence signals can be obtained with narrow emission peak and high signal-to-noise ratio after penetrating phantom tissue. This work brings a powerful idea for cutting-edge applications of intelligent optical materials, such as in vivo information security.